We propose new solid-state rotational-echo double-resonance (REDOR) NMR experiments to characterize the complexes of four proteins: (i) lumazine synthase, (ii) polymerized tubulin, (iii) L-selectin, and (iv) uracil DNA gIycosylase. The systems to be examined are not suited to anaIysis by solution-state NMR and are too large for the total-structure solid-state methods currently under development. Stable-isotope labeling will be used for both the protein and its ligands whenever possible. For lumazine synthase and polymerized tubulin, some relatively low-resolution electron or x-ray diffraction data are avaiIable. For these systems, we will start with the best available structure from diffraction analysis and then use molecular modeling, restrained by a few accurate REDOR distances, to refine or enhance the structure of the liganded binding site. Our goal in these two systems is to define the conformation of the bound ligands with sufficient accuracy that new drug molecules can be developed that have improved bioactivity. L-selectin has low-affinity binding but is still highly selective. The selectivity appears to arise from multivalency; that is, the L-selectin binds ligands using multiple interactions at the same time. These complexes do not crystallize. We propose to use REDOR to determine quantitatively the degree of multi-valency of L-selectin in binding to a natural glycoprotein substrate. This basic information could be crucial in the development of anti-inflammatory drugs. The DNA damage and repair enzyme uracil DNA glycosylase is thought to achieve its specificity by the recognition of the shapes of certain types of molecular defects. This idea will be tested by the REDOR characterization of stable-isotope labeled (non-crystallizable) mimics of transient protein-DNA complexes thought to lie along the recognition pathway. [unreadable] [unreadable]